COGNITION

Cognition 70 (1999) 211240

Acquired theory of mind impairments following stroke

Francesca Happe a ,*, Hiram Brownell b, Ellen Winner c
Social, Genetic and Developmental Psychiatry Research Centre, Institute of Psychiatry, London, UK b Department of Psychology, Boston College, Aphasia Research Center, Department of Neurology, Boston University School of Medicine, Boston, MA, USA c Department of Psychology, Boston College, Boston, MA, USA Received 16 September 1998; accepted 15 January 1999
a

Abstract The ability to attribute thoughts and feelings to self and others (theory of mind) has been hypothesised to have an innate neural basis and a dedicated cognitive mechanism. Evidence in favour of this proposal has come from autism; a brain-based developmental disorder which appears to be characterised by impaired theory of mind, despite sometimes good general reasoning skills/IQ. To date no case of specic acquired theory of mind impairment has been reported. The present study examined theory of mind in adults who had suffered right hemisphere stroke, a group known to show pragmatic and social difculties. In one study using story materials and two using cartoons, patients understanding of materials requiring attribution of mental states (e.g. ignorance, false belief) was signicantly worse than their understanding of non-mental control materials. Data from healthy elderly subjects, and a small group of left hemisphere patients (who received the tasks in modied form), suggest that this impairment on mental state tasks is not a function of task difculty. The ndings support the notion of a dedicated cognitive system for theory of mind, and suggest a role for the healthy right hemisphere in the attribution of mental states. 1999 Elsevier Science B.V. All rights reserved. Keywords: Theory of mind; Autism; Right Hemisphere stroke

1. Introduction Interest in the brain basis of social understanding has increased in recent years,
* Corresponding author. 0010-0277/99/$ - see front matter 1999 Elsevier Science B.V. All rights reserved. PII: S0010 -0 277(99)00005 -0

212

et al. / Cognition 70 (1999) 211240 F. Happe

with the convergence of evolutionary theories, primate studies, explorations of normal and abnormal development, and brain imaging investigations. Much of this work concerns a specic element of social competence: the ability to attribute mental states (e.g. beliefs and desires) in order to explain and predict peoples behaviour. This ability, termed theory of mind, normally develops in the preschool years, although there is some debate concerning the precise age at which a child should be credited with an understanding of others mental states, according to the type of task or real-life observational measure used (see e.g. Astington et al., 1988; Carruthers and Smith, 1996). It has been suggested that the childs development of theory of mind rests on an innately predisposed, dedicated cognitive mechanism, perhaps even a module (Fodor, 1992; Leslie and Roth, 1993). Evidence put forward in favour of such a notion includes the nding that young children develop understanding of beliefs, desires and false beliefs in a fairly predictable order, and over a surprisingly uniform time span, and that children from very different cultures appear to develop understanding of beliefs at similar rates and arrive at a similar belief-desire psychology (Avis and Harris, 1991). Functional brain imaging studies have shown frontal brain regions which are specically more active during theory of mind tasks than during control tasks - although there is, as yet, no agreement as to the exact regions involved (Baron-Cohen et al., 1994; Fletcher et al., 1995; Goel et al., 1995). Those in favour of a dedicated cognitive mechanism, or module, underlying theory of mind have pointed, in particular, to evidence of dissociation in developmental disorders. Theory of mind ability appears to be dissociable from cognitive ability in other domains; children with autism may be mind-blind yet intelligent in other respects (Baron-Cohen et al., 1993), and children with Williams syndrome, it has been suggested, show intact theory of mind despite delayed theorising in other domains (Karmiloff-Smith et al., 1995). Although the argument from double dissociation to separable cognitive systems is controversial (see, for example, Dunn and Kirsner, 1988; Goldberg, 1995; Plaut, 1995), ndings of uneven cognitive proles may at least provide constraints on proposed models of functional architecture. The dissolution of cognitive abilities, like the development of those abilities, has the potential to reveal distinct and dedicated modular systems, although in this case dissociation can only suggest differentiation in the mature cognitive architecture (perhaps an emergent property; Karmiloff-Smith, 1993), and is silent on the question of intrinsic or innate modularity (Goldberg, 1995). Can theory of mind be specically knocked out in a previously normal adult, leaving other reasoning intact? This question has not yet been answered, although there is increasing interest in the social and emotional decits which follow brain injury. What are the candidate groups for acquired theory of mind decits? One group might be individuals with frontal lobe lesions, who appear to show abnormalities of social and emotional functioning (Damasio et al., 1990; Saver and Damasio, 1991). A second possible group are patients with acquired right hemisphere damage (RHD), and it was this group that formed the focus for the work reported here. Adults with acquired damage to the right hemisphere are of interest because their

 et al. / Cognition 70 (1999) 211240 F. Happe Table 1 Social and communicative functions impaired in RHD and Autism Impaired function Indirect requests Metaphor Irony RHD reference Hirst et al., 1984 Foldi, 1987; Stemmer et al., 1994 Brownell et al., 1984 Van Lancker and Kempler, 1987 Kaplan et al., 1990; Brownell et al., 1992 Winner et al., 1998 Bihrle et al., 1986 Dagge and Hartje, 1985 Cicone et al., 1980; DeKosky et al., 1980 Borod et al., 1985; Borod and Koff, 1990; Martin et al., 1990 Bowers et al., 1985; Borod, 1993 Adolphs et al., 1996; Mandal et al., 1996 Tucker et al., 1977; Gorelick and Ross, 1987 Weintraub et al., 1981; Heilman et al., 1984 Wapner et al., 1981; Joanette et al., 1986 Davis et al., 1997 Autism reference Paul and Cohen, 1985 Happe , 1993 Ozonoff and Miller, 1996 Happe , 1993

213

Humour Emotion Cross modal matching Facial affect production

Ozonoff and Miller, 1996

Hobson et al., 1988; Loveland et al., 1995 Yirmiya et al., 1989; Loveland et al., 1994 Hobson, 1986a,b

Facial affect comprehension Prosody Production Comprehension Cohesive discourse Production

Baltaxe, 1984; Fine et al., 1991 Van Lancker et al., 1989

Judgement Discourse inference Integration of information Verbal Picture sequences Social behaviour Personal reference

Rehak et al., 1992 Brownell et al., 1986, 1992; Delis et al., 1983 Hough, 1990; Benowitz et al., 1990 Huber and Gleber, 1982; Benowitz et al., 1990 Gardner, 1975; Van Lancker, 1991 Brownell et al., 1997

Baltaxe and DAngiola, 1992; Baltaxe et al., 1995 Fine et al., 1994; Paul and Cohen, 1985 Surian et al., 1996 Ozonoff and Miller, 1996

Happe , 1997 Baron-Cohen et al., 1986 Frith, 1989; Borden and Ollendick, 1994 Lee et al., 1994

social and communicative impairments appear to resemble those of high-functioning people with autism (see Table 1, which includes representative, though not exhaustive, references)1. In autism, these decits have been linked to theory of mind impairment. For example, Happe (1993) found a close relation between ability
Of course, the effects of acquired and developmental disorders are dissimilar in many ways, and people with right hemisphere lesions do not resemble people with autism in other important respects.
1

214

et al. / Cognition 70 (1999) 211240 F. Happe

to process non-literal utterances (metaphor and irony) and theory of mind task performance in people with autism (and normal young children). People with autism have been shown to do poorly on pragmatic and discourse tasks sensitive to right hemisphere brain damage (Ozonoff and Miller, 1996). While it is clear that similar patterns of decit at the behavioural level do not necessarily result from similar underlying cognitive impairments, the ndings summarised in Table 1 suggest that RHD individuals might be a likely candidate group for acquired theory of mind decits. A number of authors have discussed the cognitive decits which follow RHD, many using descriptive terms (e.g. failure to assume varied perspectives, problems of pragmatics) which t well with the present hypothesis of impaired theory of mind. In addition, a number of distinct, alternative accounts have been proposed, against which the present hypothesis competes (see McDonald, 1993, for review). Heilman and colleagues (e.g. Heilman et al., 1984) and others (e.g. Ross, 1981) propose primary affective abnormalities following RHD, and suggest that difculties expressing and recognising emotion are sufcient to account for many of the social and communicative impairments recorded. Alternatively, a generalised integration decit has been proposed (Benowitz et al., 1990), which might link the communicative problems to the well-documented visuo-spatial construction decits in RHD, through a failure to integrate information in context. The present study tested the hypothesis that people with RHD show a decit in theory of mind in the context of otherwise intact reasoning skills. Such a selective decit would offer strong support for the notion of a dedicated cognitive mechanism for theory of mind, and suggest a theory of mind account for the social and communicative impairments associated with RHD.

2. Study 1: Theory of mind following right hemisphere stroke 2.1. Participants The right hemisphere damaged (RHD) group was recruited through the Aphasia Research Center of the Department of Neurology, Boston University School of Medicine and Braintree Hospital. Eight participants (ve male, three female) had taken part in research at the Aphasia Research Center for a number of years, and were between 7 and 23 years (mean 10 years) post-cerebro-vascular accident (CVA). An additional six participants (all female) had suffered more recent CVAs, between 4 and 9 months prior to testing (mean 5 months). For the total RHD group (ve men, nine women) age ranged from 51 to 75 (mean 64 years) and years of education was between 10 and 19 years (mean 13.4). An approximate IQ estimate, using the Quick Test (Ammons and Ammons, 1962), showed the group to range in ability level from 87 to 120 (mean 101). For all participants, unilateral lesion in the right hemisphere had been conrmed by CT or MRI scan. More precise localisation information was difcult to obtain, however, since scans had in most cases been performed very soon after the CVA (for

 et al. / Cognition 70 (1999) 211240 F. Happe

215

clinical purposes, to establish cause), at too early a stage in the illness to reveal the nal pattern of tissue loss. Patient notes in nine cases described a CVA in the middle cerebral artery distribution (MCA). Notes in the other cases described a CVA in the parietal, fronto-temporal or temporal regions, or (in one case) in the anterior communicating artery distribution. Lesion details, as available from hospital records, are shown in Appendix A. All participants were right-handed, had English as their rst language, and were free of additional diagnoses (past/present psychiatric disorder, developmental/learning disabilities, drug/alcohol abuse). None of the RHD group showed neglect during reading or picture description tasks at time of test. The control group consisted of 19 healthy elderly individuals (10 women, nine men) recruited from a subject pool maintained by the Aphasia Research Center of the Department of Neurology, Boston University School of Medicine. All were right-handed individuals, aged between 61 and 80 years (mean age 73), and free of past or present psychiatric diagnoses, developmental or learning disabilities, medical illness and drug or alcohol abuse. IQ was not assessed in this group. Years of education ranged from 12 to 18 (mean 14.6 years). For all participants English was the rst or joint rst language. Comparison of participant characteristics in the RHD and Control groups showed that the RHD participants were signicantly younger than the controls (t = 3.45, d.f. 31, P = 0.002), but did not differ in years of education (t = 1.59, d.f. 31, P = 0.121). Data from this group are also reported in a study comparing theory of mind in old and young adults (Happe et al., 1998). 2.2. Task 1: Story comprehension 2.2.1. Materials Sixteen short passages, each followed by a test question, were used. These materials were adapted from a study of theory of mind in autism (Happe , 1994), and subsequently used in two functional imaging studies of this ability (Fletcher et al., 1995; Happe et al., 1996). The story passages were of two types; theory of mind stories and non-mental stories. The theory of mind (ToM) stories concerned double bluff, mistakes, persuasion and white lies (two examples of each of these four story types). These stories were followed by questions requiring an inference about the characters thoughts and feelings, in most cases an inference about the speakers/ actors intentions. The non-mental stories also involved people and the subsequent test questions also required inferences to be made, but in this case the mental states of the characters were not relevant and the inference concerned, for example, physical causation. The topics of the eight non-mental stories can be summarised as follows; setting off a burglar alarm, paying for a car by instalment versus lump sum, x-raying an elderly woman following a fall, ling a book in a library, making meringues from egg whites left over after making mayonnaise, the role of weather conditions in determining the outcome of air and land battles, identifying the most likely location for reading glasses to have been mislaid, and buying multi-packs rather than single items while shopping. Examples of the two types of story can be seen in Appendix B.

216

et al. / Cognition 70 (1999) 211240 F. Happe

2.2.2. Procedure All participants were tested individually following the same procedure. Participants were told that they would be shown a number of short passages and that they were to read each passage silently until they had understood it, at which point they should turn the page for the test question, and tell their answer to the experimenter. Participants were instructed that once they had turned the page for the test question they were not to turn back to the passage, and so they were encouraged to spend as long as necessary studying the passage before turning over. The theory of mind and non-mental stories were blocked and presented in counterbalanced order. Before the rst story a practice story was given. Time to read each story (before turning the page) was recorded, and participants answers to the test questions were noted. These answers were later rated according to a standardised scoring scheme, with good agreement (85%, with disagreements resolved on discussion) from a second rater blind to subject group and hypothesis. Answers were scored 0, 1, or 2, with 2 being credited for a full and explicitly correct answer and one for a partial or implicit answer. Examples of scoring criteria for the stories are given in Appendix B. 2.2.3. Results Results from the group of elderly Controls have been reported previously with relation to age effects, in comparison with young adults (Happe et al., 1998). The Control and RHD groups scores on the test questions, and time to read the passages, are shown in Table 2. A repeated measures ANOVA comparing scores for the two groups by story type, showed a main effect of group (F(1,31) = 10.4, P = 0.003) but not of type. There was a signicant interaction of group by type (F(1,31) = 20.7, P = 0.000).2 This was due to RHD participants scoring signicantly less well than controls on ToM stories (t = 4.5, d.f. 15.52, P = 0.000), but not on non-mental stories (t = 0.1, d.f. 31, P = 0.92). In addition, the RHD group scored signicantly less well on ToM stories than on non-mental stories (t = 2.13 d.f. 13, P = 0.053), while the Control group
Table 2 Results for RHD and control groups by story condition (means SD) Group Controls (n = 19) RHD (n = 14)a
a

ToM stories Score (max = 16) Time (s) Score (max = 16) Time (s) 14.9 27.2 10.6 38.1 1.2 5.9 3.4 13.1

Control stories 12.4 33.9 12.5 39.8 2.2 8.3 2.2 12.0

Times available from 12 subjects only.

2 Data were treated as representing interval level measurement in order to facilitate statistical analysis. The critical interaction of group by condition was corroborated using non-parametric analysis of frequencies in the form of 2x2 contingency tables. The groups differed in numbers of subjects who did versus did not get a zero score on the ToM stories (x2 = 10.78, P 0.01), but did not differ in this respect on the non-mental stories (x2 = 0.004, P 0.1).

 et al. / Cognition 70 (1999) 211240 F. Happe

217

performed better on ToM than on non-mental stories (t = 4.97, d.f. 18, P = 0.000). Fig. 1 shows scatterplots giving each individuals explanation score, illustrating the degree of overlap between RHD and Control groups on non-mental stories and lack of overlap on ToM stories (see Appendix C for individual data). Time data from two RHD participants were missing, due to mechanical failure. Analysis of time data from the remaining 31 participants showed a main effect of group (F(1,29) = 6.33, P = 0.018), and of story type (F(1,29) = 13.68, P = 0.001). Again, a signicant interaction between group and story type was found (F(1,29) = 4.7, P = 0.038). Despite equal times for the non-mental stories, the RHD participants were slower than controls to read the ToM stories (t = 2.72, d.f. 13.88, P = 0.017). Control subjects were signicantly faster to read ToM than non-mental stories (t = 4.52, d.f. 18, P = 0.000), while the RHD group did not show any such time advantage for ToM stories (t = 1.06, d.f. 11, P = 0.31). Previous work (Happe et al., 1998) has shown that healthy elderly adults perform better on ToM stories than do young adults (mean age 22 years). Because the RHD group was signicantly younger than the control group, the story scores were reexamined to determine whether the critical interaction of group by story type was

Fig. 1. Scatterplots showing individual explanation scores on ToM and non-mental stories by group.

218

et al. / Cognition 70 (1999) 211240 F. Happe

Fig. 2. Examples of theory of mind and non-mental cartoons used in Task 2.

due to age effects. A difference score, the difference between each participants performance on the ToM stories and on the non-mental stories, formed the dependent variable in a simultaneous model multiple regression analysis. (A reliable group effect for the ToMnon-mental difference score is analogous to an interaction in ANOVA.) With both age and group included as predictor variables, the group variable was statistically signicant (Beta = 0.68, t(30) = 4.1, P = 0.000), conrming that the patients disproportionate difculty with ToM stories was signicant over and above any effect due to age differences across the two groups. The Beta (0.09) for the age variable was not reliably different from zero (t(30) = 0.57, P = 0.57). 2.3. Task 2: Single cartoons 2.3.1. Materials The second task used 12 single-frame cartoons taken from popular magazines (e.g. New Yorker). The cartoons formed two conditions: theory of mind (ToM) cartoons, in which the humour depended upon what a character mistakenly thought or did not know, and non-mental cartoons in which the humour did not involve a characters false belief or ignorance but instead involved a physical anomaly or violation of a social norm. Examples of each type of material are shown in Fig. 2: the ToM cartoon shown requires an inference about the fathers ignorance concerning the monster on the stairs, and consequent mistaken belief (that the boy is just telling a joke or riddle); while the non-mental cartoon requires an inference about prior physical events (that the small boy in the lab was an adult scientist who discovered the elixir of youth). Four cartoons in the ToM set and four in the non-

 et al. / Cognition 70 (1999) 211240 F. Happe

219

mental set showed facial expressions. Six cartoons (three ToM, three non-mental) included captions read aloud to participants.3 2.3.2. Procedure Participants were shown the cartoons in random order, one at a time, with the instruction to tell the experimenter why each was funny. Answer given, and time taken before answering, were recorded. Answers were scored according to a standard scoring scheme in which 3 was given for a full and explicit explanation, 2 for a partial/implicit explanation, and 1 for reference to relevant parts of the cartoon without further explanation. Irrelevant, incorrect or dont know answers were scored 0. Good inter-rater agreement was achieved (87%) and disagreements were resolved upon discussion. Examples of answers and their scoring are given in Appendix B. 2.3.3. Results The groups scores for cartoon explanation, and time to answer can be seen in Table 3. A repeated measures ANOVA comparing scores for the two groups by cartoon type showed a main effect of group (F(1,31) = 38.5, P = 0.000) but not of type. There was a signicant interaction of group by type (F (1,31) = 48.3, P = 0.000).4 This was due to RHD participants scoring much less well than controls on ToM cartoons (t = 9.38, d.f. 31, P = 0.000), but only marginally less well on non-mental cartoons (t = 1.98, d.f. 31, P = 0.06). Fig. 3 illustrates these group differences with scatterplots showing each individuals explanation score (see also Appendix C). In addition, the RHD group scored signicantly less well on ToM cartoons than on non-mental cartoons (t = 4.5, d.f. 13, P = 0.001), while the control group performed better on ToM than on non-mental cartoons (t = 5.3, d.f. 18, P = 0.000).

Table 3 Results from Single Cartoons task: mean SD score for explanations and time per item Group Controls (n = 19) RHD (n = 14)a Score (max = 18) Time (s) Mean (max = 18) Time (s) ToM Cartoons 15.3 12.2 7.1 17.9 1.9 5.1 3.1 7.3 Non-mental Cartoons 12.7 13.3 10.7 15.5 2.9 4.7 2.7 4.7

3 One of the 6 ToM cartoons contained crucial information on the far left side of the drawing, but this did not differ from the other cartoons in scores obtained by the RHD group - supporting the impression that no participant was suffering from neglect at time of test.

The critical interaction of group by condition was corroborated using non-parametric analysis of frequencies in the form of 2x2 contingency tables. When participants were classied by group (Control, RHD) there was a signicant effect for ToM cartoons (x2 = 19.77, P 0.001) but not for non-mental cartoons (x2 = 0.74, P 0.1), with relation to numbers of participants who did or did not receive more than one zero score.

220

et al. / Cognition 70 (1999) 211240 F. Happe

Fig. 3. Scatterplots showing individual explanation scores on ToM and non-mental cartoons by group.

Time data from one RHD participant for the non-mental cartoons were missing, due to mechanical failure. Analysis of time data from the remaining 32 participants showed a main effect of group (F(1,30) = 4.59, P = 0.040), but not of cartoon type. Again, a signicant interaction between group and cartoon type was found (F(1,30) = 8.37, P = 0.007). Despite equal times for the non-mental cartoons, the RHD participants were slower than controls on the ToM cartoons (t = 2.65, d.f. 30, P = 0.013). Because the RHD and control groups differed in mean age, the explanation scores were re-examined to determine whether the critical interaction of group by cartoon type was due to age effects. A difference score, the difference between each participants performance on the ToM cartoons and on the non-mental cartoons, formed the dependent variable in a simultaneous model multiple regression analysis. With both age and group included as predictor variables, the group variable was statistically signicant (Beta = 0.8, t(30) = 5.9, P = 0.000), conrming that the patients difculty with ToM cartoons was signicant over and above any effect due to age differences across the two groups. The Beta (0.04) for the age variable was not reliably different from zero (t(30) = 0.33, P = 0.74). A similar multiple regression analysis was performed for time data, using the difference

 et al. / Cognition 70 (1999) 211240 F. Happe

221

between times for ToM and times for non-mental cartoons as the dependent measure. As above, group was a signicant predictor of time difference (Beta = 0.56, t(29) = 2.95, P = 0.006), and the effect of age was not signicant (Beta = 0.18, t(29) = 0.93, P = 0.36). 2.4. Task 3: Cartoon pairs 2.4.1. Materials A second set of cartoons was selected from the same sources as those used in Task 2, and fell into the same two categories: ToM and non-mental cartoons. Five ToM

Fig. 4. Examples of theory of mind and non-mental cartoon pairs used in Task 3. Reproduced by kind permission of the artists.

222

et al. / Cognition 70 (1999) 211240 F. Happe

and ve non-mental cartoons, all non-verbal (without captions or speech), were used for this study. Three cartoons in the ToM set and three in the non-mental set showed facial expressions. For each cartoon, a matching stimulus was created by making a copy of the cartoon with the key humorous element replaced. Fig. 4 shows an example of the ToM and non-mental cartoon pairs. In each case the element was replaced (not simply omitted), so that the altered versions of ToM cartoons no longer supported an inference of false belief/ignorance, and the altered versions of nonmental cartoons no longer supported an inference of physical impossibility/convention-violation. 2.4.2. Procedure Cartoon pairs (original and altered) were presented side by side (left-right order counterbalanced). Participants were told that only one cartoon in each pair was funny, and were asked to point to the funny one. Choice was recorded as correct, incorrect or dont know. Time to point was recorded. Participants were then asked to explain why the cartoon was funny. When a participant refused to pick one cartoon from the pair, the correct (unaltered) cartoon was indicated while the explanation question was asked. Answers were recorded and scored according to the same scheme used in Study 1 (see Appendix B for examples). 2.4.3. Results The results are shown in Table 3. As can be seen, the RHD group again performed less well than the Control group on ToM cartoons. A repeated measures ANOVA for number of correct cartoon choices showed a signicant main effect of group (F(1,31) = 11.18, P = 0.000) but not of cartoon type, and a signicant interaction of group by type (F(1,31) = 8.6, P = 0.006). This interaction was due to Controls selecting a greater number of correct cartoons than did the RHD participants, when presented with ToM pairs (t = 6.43, d.f. 31, P = 0.000) but not when choosing from non-mental cartoon pairs (t = 1.38, d.f. 31, P = 0.18). For control participants, choosing the correct cartoon from ToM pairs was easier than choosing from nonmental pairs (t = 3.31, d.f. 18, P = 0.004), while in the RHD group performance across the two types did not differ signicantly (t = 1.0, d.f. 13, P = 0.34). Because the data contained ceiling effects that might affect the interpretability of ANOVA,
Table 4 Results from Cartoon Pairs task; mean SD correct choices, explanation scores, and times per cartoon pair Group Controls (n = 19) RHD (n = 14) No. correct choices (max = 5) Score (max = 15) Time (s) No. correct choices (max = 5) Score (max = 15) Time (s) ToM Cartoons 4.7 12.7 17.9 3.4 7.4 23.2 0.5 1.5 6.4 0.6 2.4 9.3 Non-mental Cartoons 4.0 10.3 24.7 3.6 9.5 21.8 0.8 2.3 9.2 0.9 2.7 8.7

 et al. / Cognition 70 (1999) 211240 F. Happe

223

Fig. 5. Scatterplots showing individual explanation scores on ToM and non-mental cartoon pairs by group.

the critical group by task interaction was examined using non-parametric tests. Analysis of frequencies in the form of 2x2 contingency tables conrmed the interaction of group by cartoon type. The numbers of subjects who ever (versus never) chose the wrong cartoon differed between groups for the ToM pairs (x2 = 15.8, P 0.001) but did not differ for the non-mental pairs (x2 = 1.3, P 0.1) (see Table 4). A repeated measures ANOVA for explanation scores showed a signicant main effect of group (F(1,31) = 19.94, P = 0.000) but not of type. In addition, there was a signicant group by cartoon type interaction (F(1,31) = 33.3, P = 0.000)5. The groups did not differ in their scores for the non-mental cartoon pairs, but the
5 The critical interaction of group by condition was corroborated using non-parametric analysis of frequencies in the form of 2 2 contingency tables. The groups differed signicantly in the number of participants who did versus did not get zero scores on the ToM cartoons (x2 = 18.9, P 0.001), but on the non-mental cartoon pairs there was no signicant difference (x2 = 0.02, P 0.1).

224

et al. / Cognition 70 (1999) 211240 F. Happe

RHD group was signicantly worse than the Control group on the ToM cartoon pairs (t = 7.8, d.f. 31, P = 0.000). Fig. 5 illustrates this group difference with scatterplots showing each individuals explanation score (see also Appendix C). Controls were signicantly better at explaining the ToM cartoons than the non-mental cartoons (t = 4.38, d.f. 18, P = 0.000). By contrast, the RHD group scored signicantly worse on ToM than on non-mental cartoons (t = 4.08, d.f. 13, P = 0.001). Although it was possible to give a high scoring explanation following an incorrect cartoon choice, these two measures are clearly not independent. Number of correct cartoon choices correlated with score for explanations in both groups (r values from 0.57 to 0.74, P 0.05). To examine whether the RHD groups poor performance on ToM cartoon explanation was the result of incorrect cartoon choice alone, a mean explanation score was calculated for just those cartoon pairs where the correct cartoon was selected as the funny one. The same group by task interaction was found (F(1,31) = 8.42, P = 0.007), with RHD patients having lower mean explanation scores than Controls for ToM (t = 4.79, d.f. 31, P = 0.000) but not for nonmental cartoons (t = 1.30, d.f. 31, P = 0.202). Thus, even with accuracy of choice taken into account, RHD patients gave signicantly poorer explanations of ToM cartoons. As in Tasks 1 and 2, a simultaneous model multiple regression analysis, in which age and group were entered as predictor variables and the difference between ToM and non-mental scores formed the dependent variable, conrmed that age differences did not account for the worse performance of the RHD group on ToM cartoons. With both age and group included as predictor variables, the group variable was statistically signicant (Beta = 0.8, t(30) = 5.2, P = 0.000), conrming that the patients disproportionate difculty with ToM cartoons was signicant over and above any effect due to age differences across the two groups. The Beta (0.1) for the age variable was not reliably different from zero (t(30) = 0.66, P = 0.51). Analysis of time data showed no signicant main effect of group or cartoon type, but a signicant interaction of group by type (F(1,31) = 8.4, P = 0.007). The groups did not differ in their times for non-mental cartoon pairs (t = 0.92, d.f. 31, P = 0.36), but there was a trend for the RHD group to perform more slowly than the Control group on ToM cartoon pairs (t = 1.93, d.f. 31, P = 0.06). In addition, Controls were signicantly faster at explaining ToM cartoons compared to non-mental cartoons (t = 3.45, d.f. 18, P = 0.003), while the RHD group did not differ in times for the two conditions (t = 0.72, d.f. 13, P = 0.48). Again, a multiple regression analysis, using difference between times in the two conditions, showed that group (Beta = 0.5, t(30) = 2.4, P = 0.023) accounted for these differences over and above age, which was not a signicant predictor (Beta = 0.005, t(30) = 0.03, P = 0.98). 2.4.4. Discussion The results of the three tasks in Study 1 suggest that people with RHD show specic impairments in understanding stories and cartoons that require mental state attribution. This difculty does not appear to be the result of greater general task difculty, since the Control group found the theory of mind materials easier than the non-mental materials in all three tasks. This latter nding ts a previous report of

 et al. / Cognition 70 (1999) 211240 F. Happe

225

superior theory of mind in old versus young people (Happe et al., 1998). Although the RHD group was somewhat younger than the Control group, age differences did not account for the present pattern of results. It would be difcult to explain the present ndings in terms of alternative theories of cognitive decits following RHD. Inference beyond the information given was a key requirement in both the ToM and non-mental stories and cartoons. Demand for recognition of emotional expression and processing of facial information were balanced across the ToM and non-mental cartoon materials. Integration of visual information was required to understand both cartoon types. As reported in previous studies (e.g. Gillikin and Derks, 1991) the RHD patients occasionally gave answers that focused on details or reected a misperception of some aspect of the visual stimuli (30 of the total 308 responses in tasks 2 and 3). However, these abnormalities were noted in answers for both types of cartoon (17 of 154 responses to ToM cartoons, 13 of 154 responses to non-mental cartoons). The relatively good performance of the RHD group on the non-mental cartoons and stories, then, rules out a number of alternative explanations for their failure on the ToM tasks. The possible role of gender, age at stroke, and time elapsed since stroke, in explaining the range of performance seen (as illustrated in Figs. 1, 3, and 5) could not be explored here due to small numbers. By chance, the individuals in the group who had suffered recent CVAs happened to be female, and so gender and time elapsed since stroke were confounded. Understanding individual differences in the sequelae of brain damage, in relation to lesion sites, is clearly an important aim for future research.

3. Study 2: Theory of mind following left hemisphere stroke Study 1 compared the performance of RHD patients with that of healthy elderly people. Is it possible that ToM impairments are not specic to RHD, but that ToM tasks are simply especially sensitive to the cognitive decits that follow any type of brain damage? Study 2 attempted to address this question by examining ToM in a small group of patients with left hemisphere damage. 3.1. Participants Five participants (four men and one woman) who had suffered left CVA took part in Study 2. All had unilateral left hemisphere lesions conrmed on CT or MRI scan, and had been assessed at the Aphasia Research Center. All had received the diagnosis of Brocas aphasia, had lesion sites consistent with this characterisation, and currently had good comprehension skills. Patient notes described lesions involving, in addition to Brocas area, the left dorso-lateral prefrontal cortex, frontal horn, frontal operculum, and MCA distribution. They ranged in age from 54 to 80 (mean 67 years), and had between 8 and 16 years of education (mean 12.6). Participants ranged from 12 months to 21 years post-CVA (mean 9 years). Participant details are shown in Appendix A.

226

et al. / Cognition 70 (1999) 211240 F. Happe

3.2. Materials The story and cartoon materials devised for Tasks 1, 2 and 3 were used. Since the LHD participants all had expressive aphasia, and proved unable to answer open-ended questions, a forced-choice answer format was used for the explanation questions. The three explanations offered (in counterbalanced order) were (1) the correct inference, (2) an incorrect/inappropriate inference, (3) verbatim information from the story/reference to a physical aspect of the cartoon. These answers were based on correct and incorrect explanations offered by Control and RHD participants in Study 1. Examples are shown in Appendix B. 3.3. Task 1: Story comprehension 3.3.1. Procedure The 16 stories used in Study 1 were shown and read aloud to the LHD patients. Patients signalled when they were ready to proceed to the test question, and the story was read a second time if required. As for the RHD and Control groups, participants were not permitted to return to the story once the page had been turned for the test question. The LHD patients were then read the test question, and offered the three forced-choice answers with the instruction to Choose the best answer. Test question and answers were repeated as necessary. Answer chosen was recorded. Because of the time needed to read aloud the three answer choices, and the occasional need to repeat an answer, time to select an explanation was not considered a meaningful reection of processing time, and was not recorded. 3.3.2. Results Of a possible total score of 8, the LHD group scored a mean of 6.6 (SD 0.5) for ToM stories and 5.8 (SD 1.7) for non-mental stories. There was no sign of impairment on the ToM stories, and performance on the two story types did not differ signicantly (t = 1.21, d.f. = 4, P = 0.29). Individual data are shown in Appendix C. 3.4. Task 2: Single cartoons 3.4.1. Procedure Participants in the LHD group were shown the cartoons as described in Study 1, and were encouraged to say if possible what was funny in each cartoon. Only two participants were able to say single words that conveyed an answer for any of the cartoons. All participants were therefore offered the choice of three possible explanations (order counterbalanced) from which they were asked to choose the explanation that best ts why the cartoon is funny. Participants choice of answer was recorded. One patient was unable to complete this task due to time constraints.

 et al. / Cognition 70 (1999) 211240 F. Happe

227

3.4.2. Results The LHD patients showed no sign of greater difculty with the ToM cartoons, and indeed performed better on ToM than on non-mental materials (see Appendix C for individual data). The mean number of correct answer choices (maximum 6) was 5 (SD 0.8) for ToM and 3.3 (SD 0.9) for non-mental cartoons, showing a signicant advantage on ToM cartoons (t = 3.66, d.f. = 3, P = 0.035). 3.5. Task 3: Cartoon pairs 3.5.1. Procedure Participants were shown the cartoon pairs (as described in Study 1 above) and asked to indicate which of each pair was the funny one. Time to choose was recorded. This response measure was exactly as in Study 1. In addition, all participants were asked Why is that one funny?, and offered the choice of three possible explanations (order counterbalanced), which fell into the same three categories described above (correct inference, incorrect inference, physical reference). Choice of explanation was recorded.

Fig. 6. Scatterplots showing individual cartoon choice scores on ToM and non-mental cartoon pairs by group (RHD, LHD and Control).

228

et al. / Cognition 70 (1999) 211240 F. Happe

3.5.2. Results The LHD patients performed equally well on the two types of cartoons (see Appendix C for individual data). There was no signicant difference in the number of correct cartoon choices (maximum 5); 4.8 (SD 0.5) for ToM pairs and 4.2 (SD 0.8) for non-mental cartoon pairs. Selection of the funny cartoon from each pair is a task directly comparable across the three groups tested. Comparing the LHD group with the RHD and normal Control groups of Study 1, a repeated measures ANOVA for group (three levels) by condition (two levels) showed a signicant main effect of group (F (2,35) = 9.93, P = 0.000), and of condition type (F(1,35) = 4.7, P = 0.04) on number of correct cartoon choices. In addition, there was a signicant interaction of group by condition (F(2,35) = 4.55, P = 0.02). One-way ANOVA with post hoc Tukey tests showed that this was due to the RHD group making signicantly fewer correct cartoon choices than both Control and LHD groups in the ToM condition only (F(2,35) = 24.6, P = 0.000; between groups Tukey P 0.05). This group difference is illustrated in Fig. 6 with scatterplots showing individual data for cartoon choice in the RHD, LHD and Control groups. Since ceiling effects threaten the validity of this ANOVA, this interaction was also explored with non-parametric tests. Analysis of frequencies showed that numbers of participants making at least one wrong cartoon choice did not differ in the LHD and Control groups for either condition (Fisher, P 0.1). By contrast, the number of participants making incorrect choices was signicantly higher in the RHD group than the LHD in the ToM condition (Fisher, P 0.005), but did not differ in the non-mental condition (P 0.1). Thus the RHD group performed less well on the ToM cartoons (but not on the non-mental cartoons) in comparison to both healthy controls and LHD patients. Time to pick the funny cartoon from the pairs presented is also a measure of performance directly comparable with Study 1. Time data were available from 4 of the LHD participants, and mean time to choose the cartoon was 13.9 s (SD 1.5 s) for ToM pairs, and 22.5 s (SD 5.7 s) for non-mental cartoons. LHD participants, like the Controls in Study 1, were signicantly faster to pick the target cartoon in the ToM condition (t = 3.23, d.f. 4, P = 0.032). Comparison of the LHD, RHD and Control groups times to pick cartoons from the ToM and non-mental pairs showed a signicant group by task interaction (F(2,34) = 5.10, P = 0.012). RHD patients took longer than LHD patients to pick the funny cartoon in the ToM condition (t = 3.57, d.f. 14.9, P = 0.003) but not in the non-mental condition (t = 0.16, d.f. 16, P = 0.88). Mean score for explanation (selection from three answers, maximum score 5) was 4.6 (SD 0.6) correct for ToM cartoon pairs and 4.2 (SD 0.8) for non-mental cartoon pairs, with no signicant difference between the two conditions (t = 1.00, d.f. 4, P = 0.37). 3.5.3. Discussion The LHD patients showed no sign of greater difculty on the ToM materials, compared with the non-mental tasks. This suggests that acquired damage to the left

 et al. / Cognition 70 (1999) 211240 F. Happe

229

hemisphere does not necessarily impair the ability to attribute mental states. The impairment on ToM tasks shown in Study 1 and 2 does seem, then, to be specic to right hemisphere damage. One limitation of the present study is that the RHD participants did not receive the forced-choice explanation task (since they had already been given the cartoon tasks in open question form). It remains possible, then, that the RHD participants would do better on the tasks, and specically on the ToM tasks, in this forced-choice mode. However, the fact that the foil answers used in the forced-choice format were taken from spontaneous answers given by RHD participants, suggests that such an improvement should not be expected. More importantly, two measures from the cartoon pairs task (choice of cartoon and time to choose cartoon) allow direct comparison bet-ween performance in the LHD group and in the RHD and Control groups. Results from these measures mirror those from the forced-choice explanation measures, and show signicant group differences, with RHD but not LHD patients performing worse than healthy controls on ToM materials only.

4. General discussion The present ndings suggest that the well-documented social and communication decits which commonly follow right hemisphere stroke may be due to acquired impairments in theory of mind. Before discussing the possible implications, however, it is necessary to mention the limitations of the present studies. The lack of detailed scan data for the patients, unfortunately prevents an exploration of the relation between specic lesion sites and theory of mind (dis)ability. In addition, it leaves open the possibility that individuals in the RHD group had larger or more serious lesions than those in the LHD group, although the equivalent performance of the groups on the non-mental materials makes this less likely. The ideal comparison group for the RHD patients would be individuals with LH lesions but intact language. However, the rarity of such patients, and the likelihood that such lesions would be small in relation to the RH groups damage, led to the inclusion in the present case of aphasic individuals. This necessitated the modication of some, but not all, of the performance measures. It remains possible, then, that the relative advantage of the LHD group over the RHD group in Tasks 1 and 2 is due to these modications. However, performance measures (choice of cartoon and time to choose) on Task 3, the cartoon pairs task, were not modied for the LHD group, and still showed signicantly better performance in LHD than RHD patients. In sum, our ndings do appear to show that ToM can be selectively impaired following stroke, but require replication with a larger group of RHD patients with more clearly mapped lesions, and comparison groups with damage in other brain regions. Future case-studies with detailed lesion information should prove especially informative. With these limitations in mind, we turn to the implications of the present preliminary ndings. The ndings suggest a role for the healthy right hemisphere in normal adult theory of mind. This ts with ndings by Winner et al. (1998), who found im-

230

et al. / Cognition 70 (1999) 211240 F. Happe

paired discrimination of jokes versus lies in RHD patients, and attributed this to problems in understanding speakers intentions. Siegal et al. (1996) have also reported failure on (very simple) theory of mind tasks in RHD groups. However, the explanation they offer for this failure, in terms of misinterpretation of experimenters questions, would not appear to account for the present pattern of data. The present ndings implicating right hemisphere processes in theory of mind are consistent with some of the data emerging from functional brain imaging studies of ToM. Baron-Cohen et al. (1994) found increased activation in right orbito-frontal regions when participants were asked to identify words that had to do with the mind (e.g. think, dream, plan). Two other functional imaging studies, however, reported greater activation (during ToM versus non-mental tasks) in frontal regions of the left hemisphere (Fletcher et al., 1995; Goel et al., 1995). In the present study, LHD patients showed no evidence of ToM impairments; however, none of these patients had lesions to the key frontal areas implicated in functional imaging studies of ToM. The relationship between imaging ndings and neurological decits is an interesting one; while left frontal regions may play key roles in the neural circuits underlying ToM, there is no reason to think that the same circuit could not be disrupted by damage to other sites. Which specic brain regions or pathways are important for ToM could not be ascertained in the present study due to insufcient lesion data, but will be an interesting issue for future research. The ndings suggest that it may be fruitful to think of acquired RHD as (in some cases) a syndrome of impaired theory of mind. Impaired ability to attribute mental states to others has implications for therapy, and relevance for relatives and caregivers. Problems representing mental states may also relate to some of the other puzzling features of RHD. For example, RHD appears to be a risk factor for psychotic, atypical schizophreni-form symptoms following stroke (Rabins et al., 1991). Frith (1992) has suggested that such symptoms may be the result of abnormal mental state attribution; an anomaly in ToM leading to misattribution and misidentication of own and others thoughts and intentions. Indeed, individuals with schizophrenia also show problems understanding cartoon humour based on characters mental states (Corcoran et al., 1997). Impaired access to ones own mental states might be important in a number of intriguing clinical decits which commonly follow right (but not left) hemisphere damage. Hence anosognosia (lack of concern for handicap such as hemiplegia) has been hypothesised to result from a breakdown in monitoring of ones own action intentions (Gold et al., 1994; Heilman, 1991). Could this be part of a more general failure of mental state representation? It remains to be seen, however, whether difculties in reading other minds extend to reading own mind. Do the present ndings suggest that the right hemisphere is implicated in autism? It appears that people with RHD and people with autism have difculty with social and communicative tasks for similar reasons - problems representing the mental states of others. The present data suggest that areas in the right hemisphere are necessary on-line for the operation of normal adult theory of mind. However, this

 et al. / Cognition 70 (1999) 211240 F. Happe

231

does not allow us to conclude that right hemisphere structures are necessary in the development of theory of mind, nor that damage elsewhere in the brain cannot disrupt theory of mind on-line or developmentally. There are, however, some intriguing hints that right hemisphere functions may be relevant to understanding autism. For example, people with autism show detail-focused processing of visuospatial information and relative inability to process congural/global information (Happe , 1996; Shah and Frith, 1983, 1993), congural processing being linked to the right hemisphere (Robertson and Lamb, 1991). Autism is also characterised by repetitive, stereotyped movements and activities (DSM-IV; APA, 1994), suppression of which is believed to be the responsibility of control centres in the right hemisphere (Brugger et al., 1996). At least one brain imaging study to date has found right hemisphere abnormalities in three individuals with a high-functioning form of autism, Asperger Syndrome (McKelvey et al., 1995). In addition, there are strong behavioural similarities between autism and so-called right hemisphere learning disability, in at least some cases of which there is known right hemisphere pathology (Gross-Tsur et al., 1995; Manoach et al., 1995). However, it is important to bear in mind the complexity introduced by developmental plasticity: in some cases, early left hemisphere pathology may result in a pattern of impaired right hemisphere skills, due to cortical reorganisation (Ogden, 1989; Polster and Rapcsak, 1994). The ndings suggest that theory of mind may have a separable brain basis, and a dedicated cognitive mechanism. The brain damage suffered by the participants in the present studies presumably affected fully developed cognitive mechanisms for theory of mind. In adulthood, then, the ability to attribute mental states can be selectively impaired, suggesting that our normal adult mind-reading abilities are not simply the product of general-purpose reasoning mechanisms. It remains to be seen how the developmental picture might differ from that found here in adults; the present ndings are mute with regard to the issue of dedicated mechanisms for theory of mind acquisition, and it is possible that a process of modularisation or cognitive specialisation could result in dedicated brain mechanisms for theory of mind later in life (Karmiloff-Smith, 1993). It is clearly an important task for the future to explore the effects of brain damage at different points in the development of theory of mind.

Acknowledgements This work was undertaken while the rst author held a Human Frontiers Science Program long-term travel fellowship at Boston College. Parts of this work were also supported by grants 2R01-NS27849 and DC00081 and by a Research Grant from Boston College. We are grateful to Jennifer Balogh, Lorrie Kirchner, and the staff of Braintree Hospital for help with recruitment and testing. This paper has beneted greatly from the comments of a number of colleagues, especially Uta Frith and Chris Frith, who had the original idea to use cartoons to explore theory of mind.

232

et al. / Cognition 70 (1999) 211240 F. Happe

Appendix A. Patient information

Right hemisphere-damaged patients ID CB VF SB MC KE JM AM SO JD VA DS JH JF CL CVA description Middle cerebral artery distribution Parietal regions Frontal region extending to mid parietal Midbrain Middle cerebral artery distribution Middle cerebral artery, parietal regions Middle cerebral artery distribution Middle cerebral artery distribution Middle cerebral artery distribution Posterior frontal Frontal temporal-parietal Middle cerebral artery distribution Middle cerebral artery, temporal and parietal Middle cerebral and anterior communicating artery Sex M F M F M F M M F F F F F F Age 63 54 75 51 65 53 63 74 70 58 74 55 74 71 Time post onset 10 years 8 years 10 years 7 years 9 years 7 years 23 years 10 years 5 months 6 months 6 months 3 months 4 months 4 months Education Quick test (years) IQ estimate 16 11 11 16 19 12 16 14 12 10 14 13 12 12 108 92 96 98 102 108 108 108 100 92 90 120 87 100

Left hemisphere-damaged patients ID WF FC RD JC EM CVA Description Brocas area, lateral frontal, frontal opercullum, putamen Middle cerebral artery, Brocas area, dorso-lateral prefrontal Brocas area, insula, frontal horn, temporal lobe Middle cerebral artery, Brocas, frontal-parietal Middle cerebral artery, Brocas, parietal, posterior frontal Sex M M M F M Age 60 63 80 54 78 Time post onset 2 years 21 years 10 years 1 year 10 years Education (years) 9 16 14 16 8

Appendix B. Examples of theory of mind stories and non-mental stories Example Theory of Mind Story A burglar who has just robbed a shop is making his getaway. As he is running home, a policeman on his beat sees him drop his glove. He doesnt know the man is a burglar, he just wants to tell him he dropped his glove. But when the policeman shouts out to the burglar, Hey, you! Stop!, the burglar turns round, sees the policeman and gives himself up. He puts his hands up and admits that he did the break-in at the local shop.

 et al. / Cognition 70 (1999) 211240 F. Happe

233

Q: Why did the burglar do that? Example responses scored: 2: Because he thought the policeman knew he had robbed the shop 1: Because he thought he was caught Example Non-mental Story A burglar is about to break into a jewellers shop. He skilfully picks the lock on the shop door. Carefully he crawls under the electronic detector beam. If he breaks this beam it will set off the alarm. Quietly he opens the door of the store-room and sees the gems glittering. As he reaches out, however, he steps on something soft. He hears a screech and something small and furry runs out past him, towards the shop door. Immediately the alarm sounds. Q: Why did the alarm go off? Example responses scored: 2: Because the burglar disturbed a cat, which ran through the detector beam 1: Because something broke the beam Example answers and scoring for Single Cartoons task (see Fig. 2) ToM Cartoon 3: He thinks the kids just asking him a riddle - doesnt realise there really is a monster there! 2: He hasnt seen the thing on the stairs! 1: There really is something there with two horns and one eye Non-mental Cartoon 3: The scientist has discovered a potion to make him a kid again! 2: The scientist has made something which has made him shrink! 1: He must have shrunk! Example answers and scoring for Cartoon Pairs task (see Fig. 4) ToM Cartoon 3: Hes got each girl thinking his song is just for her 2: You cant tell which girl hes singing to 1: The guys singing and they listen Non-mental Cartoon 3: Their lines went wrong when they had to pass in the middle 2: The ladder wobbled, maybe they had a ght? 1: The painters made a mistake Example forced-choice answers offered to Aphasic participants Stories (1) Appropriate inference choice, e.g. Because he thinks the policeman knows he robbed the shop (2) Inappropriate inference choice, e.g. Because he thinks the policeman wants to give him back his glove (3) Verbatim information choice,

234

Appendix C. Individual participants performance

ID Story task ToM Score Time (max = 16) (s) RHD CB 16.00 VF 12.00 SB 10.00 MC 14.00 KE 12.00 JM 14.00 AM 14.00 SO 5.00 JD 9.00 VA 5.00 DS 11.00 JH 7.00 JF 8.00 CL 11.00 Healthy controls CS 13.00 JH 15.00 BS 13.00 SS 16.00 NW 15.00 OM 16.00 BR 16.00 JR 16.00 MB 16.00 67.20 40.60 38.60 46.40 28.00 21.20 56.00 34.40 36.90 28.60 31.40 28.00 23.00 25.50 22.20 30.40 27.50 20.50 25.50 25.90 24.10 non-ToM Score Time (max = 16) (s) 14.00 14.00 12.00 13.00 9.00 16.00 14.00 14.00 14.00 11.00 9.00 11.00 10.00 14.00 15.00 14.00 13.00 11.00 10.00 11.00 14.00 11.00 16.00 66.60 45.50 34.20 52.90 39.60 30.40 52.00 36.10 29.60 27.80 32.20 31.20 21.80 26.90 31.50 30.90 48.80 24.90 29.90 27.50 33.50 Single cartoons ToM Score Time (max = 18) (s) 14.00 8.00 6.00 11.00 3.00 8.00 9.00 6.00 3.00 9.00 4.00 5.00 6.00 7.00 13.00 15.00 16.00 17.00 16.00 17.00 15.00 15.00 17.00 10.67 18.33 16.67 17.33 30.17 9.83 12.50 25.17 15.00 10.17 17.50 11.33 32.00 18.83 13.00 17.17 5.83 7.83 7.17 6.67 14.17 14.50 19.67 non-ToM Score Time (max = 18) (s) 15.00 10.00 11.00 12.00 12.00 11.00 15.00 5.00 12.00 9.00 7.00 10.00 10.00 11.00 17.00 15.00 14.00 16.00 12.00 14.00 12.00 13.00 14.00 16.50 17.83 14.67 10.83 22.83 7.67 18.17 14.17 10.33 15.33 14.67 24.33 13.83 9.00 19.67 9.50 12.17 9.67 7.67 12.00 13.67 22.00 Cartoon pairs ToM Choice (max = 5) 4.00 4.00 4.00 3.00 3.00 4.00 4.00 2.00 3.00 4.00 3.00 3.00 3.00 4.00 5.00 5.00 5.00 5.00 5.00 4.00 5.00 5.00 4.00 Score Time (max = 15) (s) 10.00 10.00 9.00 9.00 4.00 8.00 9.00 6.00 6.00 9.00 4.00 8.00 3.00 9.00 13.00 15.00 14.00 13.00 12.00 12.00 14.00 13.00 10.00 38.00 17.80 13.80 22.60 28.60 12.40 14.80 37.00 21.60 9.60 21.00 24.80 36.60 25.60 15.00 29.60 12.40 13.60 27.80 12.40 10.40 10.60 16.40 non-ToM Choice (max = 5) 5.00 5.00 3.00 4.00 2.00 4.00 4.00 3.00 4.00 4.00 2.00 3.00 4.00 4.00 5.00 5.00 4.00 5.00 5.00 4.00 4.00 3.00 4.00 Score Time (max = 15) (s)

et al. / Cognition 70 (1999) 211240 F. Happe

14.00 12.00 11.00 12.00 5.00 9.00 14.00 8.00 7.00 8.00 9.00 8.00 7.00 9.00 13.00 15.00 10.00 11.00 9.00 10.00 10.00 6.00 10.00

23.00 16.40 18.80 20.40 26.20 13.40 19.40 40.80 10.80 18.20 15.60 15.20 31.80 34.60 17.20 45.00 21.80 17.00 24.60 20.80 16.20 26.60 49.60

JA CF SF JH BB GC CW EK MT JS Single cartoons ToM Score (max = 6) nonToM ToM Choice (max = 5) 5.00 5.00 13.20 5.00 5.00 Score (max = 5) 5.00 4.00 5.00 4.00 Time (s) 15.60 5.00 4.00 12.20 14.60 Cartoon pairs non-ToM Choice (max = 5) 4.00 22.20 5.00 3.00

14.00 16.00 16.00 13.00 13.00 16.00 16.00 15.00 14.00 14.00

21.40 17.20 32.50 26.90 37.20 27.90 30.90 32.40 24.50 41.50

14.00 15.00 13.00 7.00 10.00 14.00 14.00 11.00 12.00 11.00

26.60 27.10 48.20 32.80 43.00 32.80 48.90 40.60 30.40 37.20

17.00 18.00 14.00 11.00 12.00 16.00 13.00 16.00 15.00 18.00

6.50 7.50 18.83 9.17 14.83 13.17 8.50 11.67 11.67 23.33

15.00 15.00 9.00 6.00 7.00 11.00 12.00 13.00 11.00 15.00

6.33 11.33 18.67 9.00 12.50 17.33 12.00 18.67 11.83 20.50

4.00 5.00 5.00 4.00 4.00 4.00 5.00 5.00 5.00 5.00

12.00 15.00 13.00 12.00 10.00 12.00 13.00 12.00 15.00 11.00

10.80 15.00 18.20 24.20 15.80 19.40 14.50 30.80 20.60 22.40

5.00 4.00 4.00 3.00 3.00 4.00 4.00 3.00 3.00 5.00

12.00 10.00 9.00 8.00 6.00 12.00 12.00 10.00 9.00 13.00

12.60 27.80 29.40 21.00 19.00 20.00 22.20 30.40 23.80 24.00

Story task

ToM

nonToM

et al. / Cognition 70 (1999) 211240 F. Happe

Score (max = 8)

Score (max = 5) 3.00 5.00 5.00 4.00

Time (s) 15.00 5.00 24.20 28.60

LHD WF FC RD JC ER 5.00 5.00 4.00 6.00 4.00 4.00 4.00 4.00 3.00 2.00

7.00 7.00 6.00 7.00 6.00

8.00 6.00 3.00 6.00 6.00

235

236

et al. / Cognition 70 (1999) 211240 F. Happe

e.g. Because the policeman on his beat sees the burglar drop his glove. Cartoons (1) Appropriate humorous inference choice, e.g. The father hasnt seen the monster - he thinks his son is telling a joke! (2) Inappropriate inference choice, e.g. The little boy is telling a joke about a monster with one eye and two horns! (3) Reference to a physical detail in the cartoon choice, e.g. The little boy doesnt have a mouth, and the father has a silly face!

References
Adolphs, R., Domasio, H., Tranel, D., Domario, A.R., 1996. Cortical systems for the recognition of emotion in facial expressions. Journal of Neuroscience 16, 76787687. Ammons, R.B., Ammons, C.H., 1962. Quick Test. Psychological Test Specialists, Missoula, Mont. American Psychiatric Association, 1994. Diagnostic and Statistical Manual of Mental Disorders, 4th edn, (DSM-IV). American Psychiatric Association, Washington, DC. Astington, J.W., Harris, P.L., Olson, D.R. (Eds.), 1988. Developing Theories of Mind. Cambridge University Press, New York. Avis, J., Harris, P.L., 1991. Belief-desire reasoning among Baka children: evidence for a universal conception of mind. Child Development 62, 460467. Baltaxe, C.A., 1984. Use of contrastive stress in normal, aphasic, and autistic children. Journal of Speech and Hearing Research 27, 97105. Baltaxe, C.A., DAngiola, N., 1992. Cohesion in the disclosure interaction of autistic, specically language-impaired, and normal children. Journal of Autism and Developmental Disorders 22, 121. Baltaxe, C.A.M., Russell, A., DAngiola, N., Simmons, J.Q., 1995. Discourse cohesion in the verbal interactions of individuals diagnosed with autistic disorder or schizotypal personality disorder. Australia and New Zealand Journal of Developmental Disabilities 20, 7996. Baron-Cohen, S., Leslie, A.M., Frith, U., 1986. Mechanical, behavioural and intentional understanding of picture stories in autistic children. British Journal of Developmental Psychology 4, 113125. Baron-Cohen, S., Tager-Flusberg, H., Cohen, D.J. (Eds.), 1993. Understanding Other Minds: Perspectives From Autism. Oxford University Press, Oxford. Baron-Cohen, S., Ring, H., Moriarty, J., Schmitz, B., Costa, D., Ell, P., 1994. The brain basis of theory of mind: the role of the orbitofrontal region. British Journal of Psychiatry 165, 640649. Benowitz, L.I., Moya, K.L., Levine, D.N., 1990. Impaired verbal reasoning and constructional apraxia in subjects with right hemisphere damage. Neuropsychologia 23, 231241. Bihrle, A.M., Brownell, H.H., Powelson, J.A., Gardner, H., 1986. Comprehension of humorous and nonhumorous materials by left and right brain-damaged patients. Brain and Cognition 5, 399411. Borden, M.C., Ollendick, T.H., 1994. An examination of the validity of social subtypes in autism. Journal of Autism and Developmental Disorders 24 (1), 2337. Borod, J.C., 1993. Cerebral mechanisms underlying facial, prosodic, and lexical emotional expression: a review of neuropsychological studies and methodological issues. Special Section: Neuropsychological perspectives on components of emotional processing. Neuropsychology 7 (4), 445463. Borod, J.C., Koff, E., 1990. Lateralization for facial emotional behavior: a methodological perspective. International Journal of Psychology 25, 157177. Borod, J.C., Koff, E., Lorch, M.P., Nicholas, M., 1985. Channels of emotional expression in patients with unilateral brain damage. Archives of Neurology 42, 345348. Bowers, D., Bauer, R.M., Coslett, H.B., Heilman, K.M., 1985. Processing of faces by patients with unilateral hemisphere lesions. I. Dissociation between judgments of facial affect and facial identity. Brain and Cognition 4 (3), 258272.

 et al. / Cognition 70 (1999) 211240 F. Happe

237

Brownell, H.H., Carroll, J.J., Rehak, A., Wingeld, A., 1992. The use of pronoun anaphora and speaker mood in the interpretation of conversational utterances by right hemisphere brain-damaged patients. Brain and Language 43, 121147. Brownell, H., Pincus, D., Blum, A., Rehak, A., Winner, E., 1997. The effects of right-hemisphere brain damage on patients use of terms of personal reference. Brain and Language 57, 6079. Brownell, H.H., Potter, H.H., Bihrle, A.M., Gardner, H., 1986. Inference decits in right brain-damaged patients. Brain and Language 27, 310321. Brownell, H.H., Potter, H.H., Michelow, D., 1984. Sensitivity to lexical denotation and connotation in brain-damaged patients: a double dissociation? Brain and Language 22, 253265. Brugger, P., Monsch, A.U., Johnson, S.A., 1996. Repetitive behavior and repetition avoidance: the role of the right hemisphere. Journal of Psychiatry and Neuroscience 21, 5356. Carruthers, P., Smith, P.K., (Eds.), 1996. Theories of Theories of Mind. Cambridge University Press, Cambridge, MA. Cicone, M., Wapner, W., Gardner, H., 1980. Sensitivity to emotional expressions and situations in organic patients. Cortex 16, 145158. Corcoran, R., Cahill, C., Frith, C.D., 1997. The appreciation of visual jokes in people with schizophrenia: a study of mentalizing ability. Schizophrenia Research 24, 319327. Dagge, M., Hartje, W., 1985. Inuence of contextual complexity on the processing of cartoons by patients with unilateral lesions. Cortex 21, 607616. Damasio, A.R., Tranel, D., Damasio, H., 1990. Individuals with sociopathic behavior caused by frontal damage fail to respond autonomically to social stimuli. Behavioral Brain Research 41, 8194. Davis, G.A., ONeil-Pirozzi, T., Coon, M., 1997. Referential cohesion and logical coherence of narration after right hemisphere stroke. Brain and Language 56, 183210. DeKosky, S.T., Heilman, K.M., Bowers, D., Valenstein, E., 1980. Recognition and discrimination of emotional faces and pictures. Brain and Language 9, 206214. Delis, D.C., Wapner, W., Gardner, H., Moses, J.A., 1983. The contribution of the right hemisphere to the organization of paragraphs. Cortex 19, 4350. Dunn, J.C., Kirsner, K., 1988. Discovering functionally independent mental processes: the principle of reversed association. Psychological Review 95, 91101. Fine, J., Bartolucci, G., Ginsberg, G., Szatmari, P., 1991. The use of intonation to communicate in pervasive developmental disorders. Journal of Child Psychology and Psychiatry 32, 771782. Fine, J., Bartolucci, G., Szatmari, P., Ginsberg, G., 1994. Cohesive discourse in pervasive developmental disorders. Journal of Autism and Developmental Disorders 24, 315329. Fletcher, P.C., Happe , F., Frith, U., Baker, S.C., Dolan, R.J., Frackowiak, R.S.J., Frith, C.D., 1995. Other minds in the brain: a functional imaging study of theory of mind in story comprehension. Cognition 57, 109128. Fodor, J.A., 1992. A theory of the childs theory of mind. Cognition 44, 283296. Foldi, N.S., 1987. Appreciation of pragmatic interpretation of indirect commands: comparison of right and left hemisphere brain-damaged patients. Brain and Language 31, 88108. Frith, C.D., 1992. Cognitive Neuropsychology of Schizophrenia. Erlbaum, Hove. Frith, U., 1989. Autism: Explaining the Enigma. Blackwell, Oxford. Gardner, H., 1975. The Shattered Mind. Knopf, New York. Gillikin, L.S., Derks, P.L., 1991. Humor appreciation and mood in stroke patients. Cognitive Rehabilitation, 3035. Goel, V., Grafman, J., Sadato, N., Hallett, M., 1995. Modelling other minds. NeuroReport 6, 1741 1746. Gold, M., Adair, J.C., Jacobs, D.H., Heilman, K.M., 1994. Anosognosia for hemiplegia: an electrophysiologic investigation of the feed-forward hypothesis. Neurology 44, 18041808. Goldberg, E., 1995. Rise and fall of modular orthodoxy. Special Issue: Modularity and the brain. Journal of Clinical and Experimental Neuropsychology 17, 193208. Gross-Tsur, V., Shalev, R.S., Manor, O., Amir, N., 1995. Developmental right-hemisphere syndrome: Clinical spectrum of the nonverbal learning disability. Journal of Learning Disabilities 28, 8086. Gorelick, P.B., Ross, E.D., 1987. The aprosodias: further functional-anatomical evidence for the organi-

238

et al. / Cognition 70 (1999) 211240 F. Happe

sation of affective language in the right hemisphere. Journal of Neurology, Neurosurgery and Psychiatry 50, 553560. Happe , F.G.E., 1993. Communicative competence and theory of mind in autism: a test of relevance theory. Cognition 48, 101119. Happe , F.G.E., 1994. An advanced text of theory of mind: understanding of story characters thoughts and feelings by able autistic, mentally handicapped and normal children and adults. Journal of Autism and Developmental Disorders 24, 129154. Happe , F.G.E., 1996. Studying weak central coherence at low levels: children with autism do not succumb to visual illusions. A research note. Journal of Child Psychology and Psychiatry 357, 873877. Happe , F.G.E., 1997. Central coherence and theory of mind in autism: reading homographs in context. British Journal of Developmental Psychology 15, 112. Happe , F., Ehlers, S., Fletcher, P., Frith, U., Johansson, M., Gillberg, C., Dolan, R., Frackowiak, R., Frith, C., 1996. Theory of mind in the brain: evidence from a PET scan study of Asperger syndrome. NeuroReport 8, 197201. Happe , F., Winner, E., Brownell, H., 1988. The getting of wisdom: theory of mind in old age. Developmental Psychology 34, 358362. Heilman, K.M., 1991. Anosognosia: possible neuropsychological mechanisms. In: Prigatano, G., Schacter, D. (Eds.), Awareness of Decit After Brain Injury. Oxford University Press, New York, pp. 53 62. Heilman, K.M., Bowers, D., Speedie, L., Coslett, H.B., 1984. Comprehension of affective and nonaffective prosody. Neurology 34, 917921. Hirst, W., LeDoux, J., Stein, S., 1984. Constraints on the processing of indirect speech acts: evidence from aphasiology. Brain and Language 23, 2633. Hobson, R.P., 1986a. The autistic childs appraisal of expressions of emotion. Journal of Child Psychology and Psychiatry 27, 321342. Hobson, R.P., 1986b. The autistic childs appraisal of expressions of emotion: a further study. Journal of Child Psychology and Psychiatry 27, 671680. Hobson, R.P., Ouston, J., Lee, T., 1988. Emotion recognition in autism: coordinating faces and voices. Psychological Medicine 18, 911923. Hough, M.S., 1990. Narrative comprehension in adults with right and left hemisphere brain-damage: theme organization. Brain and Language 38, 253277. Huber, W., Gleber, J., 1982. Linguistic and nonlinguistic processing of narratives in aphasia. Brain and Language 16, 118. Joanette, Y., Goulet, P., Goulet, P., Ska, B., Nesponlous, J.-L., 1986. Informative content of narrative discourse in right-brain-damaged right-handers. Brain and Language 29, 81105. Kaplan, J.A., Brownell, H.H., Jacobs, J.R., Gardner, H., 1990. The effects of right hemisphere damage on the pragmatic interpretation of conversational remarks. Brain and Language 38, 315333. Karmiloff-Smith, A., 1993. Beyond Modularity: a Developmental Perspective on Cognitive Science. MIT Press, Cambridge MA. Karmiloff-Smith, A., Klima, E., Bellugi, U., Grant, J., Baron-Cohen, S., 1995. Is there a social module? Language, face processing and theory of mind in individuals with Williams Syndrome. Journal of Cognitive Neuroscience 7, 196208. Lee, A.H., Hobson, R.P., Chiat, S., 1994. I, you, me, and autism: an experimental study. Journal of Autism and Developmental Disorders 24, 155176. Leslie, A.M., Roth, D., 1993. What autism teaches us about metarepresentation. In: Baron-Cohen, S., Tager-Flusberg, H., Cohen, D.J. (Eds.), Understanding Other Minds: Perspectives From Autism. Oxford University Press, Oxford, pp. 83111. Loveland, K.A., Tunali-Kotoski, B., Pearson, D.A., Brelsford, K.A., Ortegon, J., Chen, R., 1994. Imitation and expression of facial affect in autism. Development and Psychopathology 6, 433444. Loveland, K.A., Tunali-Kotoski, B., Chen, R., Brelsford, K.A., Ortegon, J., Pearson, D.A., 1995. Intermodal perception of affect in persons with autism or Down syndrome. Development and Psychopathology 7, 409418.

 et al. / Cognition 70 (1999) 211240 F. Happe

239

McDonald, S., 1993. Viewing the brain sideways? Frontal versus right hemisphere explanations of nonaphasic language disorders. Aphasiology 7, 535549. McKelvey, J.R., Lambert, R., Mottron, L., Shevell, M.I., 1995. Right-hemisphere dysfunction in Aspergers syndrome. Journal of Child Neurology 10, 310314. Mandal, M.K., Mohanty, A., Pandey, R., Mohanty, S., 1996. Emotion-specic processing decit in focal brain-damaged patients. International Journal of Neuroscience 84, 8795. Manoach, D.S., Sandson, T.A., Weintraub, S., 1995. The developmental social-emotional processing disorder is associated with right hemisphere abnormalities. Neuropsychiatry, Neuropsychology, and Behavioral Neurology 8, 99105. Martin, C.C., Borod, J.C., Alpert, M., Brozgold, A.Z. et al., 1990. Spontaneous expression of facial emotion in schizophrenic and right-brain-damaged patients. Special Issue: Faces, voices, and feelings: experimental techniques and clinical implications. Journal of Communication Disorders 23, 287301. Ogden, J.A., 1989. Visuospatial and other right-hemispheric functions after long recovery periods in left-hemispherictomized subjects. Neuropsychologia 27, 765776. Ozonoff, S., Miller, J.N., 1996. An exploration of right-hemisphere contributions to the pragmatic impairments of autism. Brain and Language 52, 411434. Paul, R., Cohen, D.J., 1985. Comprehension of indirect requests in adults with autistic disorders and mental retardation. Journal of Speech and Hearing Research 28, 475479. Plaut, D.C., 1995. Double dissociation without modularity: evidence from connectionist neuropsychology. Special Issue: Modularity and the brain. Journal of Clinical and Experimental Neuropsychology 17, 291321. Polster, M.R., Rapcsak, S.Z., 1994. Hierarchical stimuli and hemispheric specialization: two case studies. Cortex 30, 487497. Rabins, P.V., Starkstein, S.E., Robinson, R.G., 1991. Risk factors for developing atypical (schizophreniform) psychosis following stroke. Journal of Neuropsychiatry and Clinical Neurosciences 3, 69. Rehak, A., Kaplan, J.A., Gardner, H., 1992. Sensitivity to conversational deviance in right-hemispheredamaged patients. Brain and Language 42, 203217. Robertson, L.C., Lamb, M.R., 1991. Neuropsychological contribution to theories of part/whole organization. Cognitive Psychology 23, 299330. Ross, E.D., 1981. The aprosodias: functional-anatomic organization of the affective components of language in the right hemisphere. Archives of Neurology 38, 561569. Saver, J.L., Damasio, A.R., 1991. Preserved access and processing of social knowledge in a patient with acquired sociopathy due to ventromedial frontal damage. Neuropsychologia 29, 12411249. Shah, A., Frith, U., 1983. An islet of ability in autistic children: a research note. Journal of Child Psychology and Psychiatry 24, 613620. Shah, A., Frith, U., 1993. Why do autistic individuals show superior performance on the Block Design task?. Journal of Child Psychology and Psychiatry 34, 13511364. Siegal, M., Carrington, J., Radel, M., 1996. Theory of mind and pragmatic understanding following right hemisphere damage. Brain and Language 53, 4050. Stemmer, B., Giroux, F., Joanette, Y., 1994. Production and evaluation of requests by right hemisphere brain damaged individuals. Brain and Language 47, 131. Surian, L., Baron-Cohen, S., Van der Lely, H., 1996. Are children with autism deaf to Gricean maxims? Cognitive Neuropsychiatry 1, 5571. Tucker, D.M., Watson, R.T., Heilman, K.M., 1977. Discrimination and evocation of affectively intoned speech in patients with right parietal disease. Neurology 27, 947958. Van Lancker, D., Cornelius, C., Kreiman, J., 1989. Recognition of emotional-prosodic meaning in speech by autistic, schizophrenic and normal children. Developmental Neuropsychology 5, 207226. Van Lancker, D.R., Kempler, D., 1987. Comprehension of familiar phrases by left- but not by righthemisphere damaged patients. Brain and Language 32, 265277. Van Lancker, D., 1991. Personal relevance and the human right hemisphere. Brain and Cognition 17, 64 92. Wapner, W., Hamby, S., Gardner, H., 1981. The role of the right hemisphere in the appreciation of complex linguistic materials. Brain and Language 14, 1533.

240

et al. / Cognition 70 (1999) 211240 F. Happe

Weintraub, S., Mesulam, M.M., Kramer, L., 1981. Disturbances in prosody. Archives of Neurology 38, 742744. Winner, E., Brownell, H., Happe , F., Blum, A., Pincus, D., 1998. Distinguishing lies from jokes: theory of mind decits and discourse interpretation in right hemisphere brain-damaged patients. Brain and Language 62, 89106. Yirmiya, N., Kasari, C., Sigman, M., Mundy, P., 1989. Facial expressions of affect in autistic, mentally retarded and normal children. Journal of Child Psychology and Psychiatry 30, 725735.